4
$\begingroup$

Considering the Jacobi theta: $\theta_3(z) = \sum_{n\in\mathbb{Z}} q^{n^2}$, we can invert $\theta_3-1$ in a small enough neighbourhood of 0.

Routine computation with Lagrange-Burmann inversion gives that the inverse have expansion starting by:

$\frac{q}{2}-\frac{q^4}{16}+\frac{q^7}{32}-\frac{q^9}{512}-\frac{11 q^{10}}{512}+\frac{13 q^{12}}{4096}+ O(q^{13})$

Is there any known closed form (or recursive formula) giving the terms of this series?

(it is very closely related to OEIS A259938 suite: https://oeis.org/A259938, corresponding to $\theta_3(q/2)$ but which doesn't seem very studied as well)

Improve this question
$\endgroup$
8
  • 3
    $\begingroup$ By closely related, one should understand related by the substitution $q \mapsto q/2$. $\endgroup$
    – F. C.
    Jul 5, 2021 at 18:00
  • 3
    $\begingroup$ Morse-Feshbach (see bottom of page) gives a sum over partitions of $n-1$ into numbers one less than a square; it's not a closed form or an elegant recurrence, but if your concern is practical evaluation then it may be of interest. $\endgroup$
    – Peter Taylor
    Jul 6, 2021 at 7:38
  • 1
    $\begingroup$ With a bit of simplification, $$A_n = n2^{-n} \sum_{s,t,u,\ldots} (-1)^{s+t+u+\cdots} \binom{n-1+s+t+u+\cdots}{s,t,u,\ldots}$$ where the sum is over non-negative integer solutions to $s + 2t + 3u + \cdots = n-1$. $\endgroup$
    – Peter Taylor
    Jul 6, 2021 at 7:47
  • 1
    $\begingroup$ To avoid confusion it would be better to call it reversion rather than inversion (the latter can be more likely understood as multiplicative inverse). $\endgroup$
    – მამუკა ჯიბლაძე
    Jul 6, 2021 at 9:18
  • 1
    $\begingroup$ Apologies, I seem to have lost a reciprocal along the way. Change $n2^{-n}$ to $n^{-1} 2^{-n}$. $\endgroup$
    – Peter Taylor
    Jul 6, 2021 at 10:13

1 Answer 1

Reset to default
3
$\begingroup$

Perhaps, the form given by Lagrange inversion theorem cannot be much simplified here. It expresses the $n$-th coefficient of a series reversion as the sum of $n-1$ values of exponential Bell polynomials.

From the practical perspective, since $\theta_3-1$ contains nonzero coefficients only at square powers, computation of the $n$-th reversion coefficient amounts to iterating over the partitions of $n-1$ into squares $>1$ decreased by $1$ (almost OEIS A243148 if we fix a number of squares). This yields the formula that Peter Taylor gave in the comments (modulo the corrections) for $n>1$:

$$A_n = \frac{1}{n!2^n} \sum_{(2^2-1)j_2 + (3^2-1)j_3 + \dots = n-1} (-1)^{j_2+j_3+\dots}\cdot \frac{(n-1+j_2+j_3+\dots)!}{j_2!j_3!\dots}.$$

Improve this answer
$\endgroup$
1
  • $\begingroup$ Seems indeed to be the best we can do. I'll accept it as an answer to the question, thanks ! $\endgroup$
    – user70925
    Jul 7, 2021 at 9:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.